海洋生物-光学算法研究──南黄海叶绿素浓度的估算

根据1995年9月中日合作南黄海现场观测资料，对南黄海生物一光学算法进行研究。首先对南黄海现场多光谱反射比数据进行因子分析，解反射比协方差矩阵的特征问题；解出特征值和对应的特征向量，求出主因子，然后建立叶绿素a浓度对主因子的多元线性回归方程，并对算法进行回报和预报检验。计算结果表明，该算法具有较高精度，同时也显示出天气条件对估算精度产生较大影响。

A STUDY ON MARINE BIO-OPTICAL ALGORITHM-ESTIMATION OF CHLOROPHYLL CONCENTRATION IN THE SOUTH YELLOW SEA

On the basis of the multi-spactral reflectanee data in the South YellowSea and the chlorophyll a concentration data determined in situ with a fluorimeterduring the "Sino-Japanese Joint Basic Research on visible light remote sensing in theYellow Sea" in September, l995, a marine bio-optical algoritm has been studied'Factor analysis has been undertaken for the multi-spectral reflectance data. Thereflectance data covariant matrix is solved for eigenvalues and their correspondingeigenvecters; the principal factors are obtained- The contributions of the first three'principal factors to the varianee of reflectance are, respectively, 66.37%, 3l.8%and 0.97%' From the fist three principal factors 99.l9% of the total informatonon the origina variables can be retrieved. A multivariate linear regression equationhas been established between the natural logarithm of the chlorophyll a coneentrationand the fist three principal factors' The result shows that the overal correlationcoefficient of the subsample reaches 0.97l3 and the root-mean-square error of thehindcasting test estimate of chlorophyll a coneentration is 0.0262 mg /m3 (the rangof the chlorophyll a: 0'l30 - 0.840 mg /m3), with a relative error of l0.06%' Therelative error for the forecasting test is l7.79%. Predicted values of chlorophyll aconeentration are greatly affeeted in cloudy and rainy days.